A new method to assess the risk of local and widespread flooding on rivers and coasts

Abstract

To date, national‐ and regional‐scale flood risk assessments have provided valuable information about the annual expected consequences of flooding, but not the exposure to widespread concurrent flooding that could have damaging consequences for people and the economy. We present a new method for flood risk assessment that accommodates the risk of widespread flooding. It is based on a statistical conditional exceedance model, which is fitted to gauged data and describes the joint probability of extreme river flows or sea levels at multiple locations. The method can be applied together with data from models for flood defence systems and economic damages to calculate a risk profile describing the probability distribution of economic losses or other consequences aggregated over a region. The method has the potential to augment national or regional risk assessments of expected annual damage with new information about the likelihoods, extent and impacts of events that could contribute to the risk.

Improving statistical models for flood risk assessment

“…As the application is an illustration, we only focus on spatial dependence and consider concurrent observations. The derived methodology can easily be extended to incorporate temporal dependence, details of this can be found in Keef et al [4] and Lamb et al [3].…”

“…The conditional exceedance model is adopted as it is able to handle the range of extremal dependence structures, including both asymptotic independence and asymptotic dependence, which are observed within the extremes of a river flow data set [3]. Previous extensions to the conditional exceedance model were developed by Keef et al [4], to account for missing data and temporal dependence and Keef et al [5], to obtain an improvemen t in the estimation of the dependence parameters under negative dependence.…”

Generation of Spatially Heterogeneous Flood Events in an Alpine Region—Adaptation and Application of a Multivariate Modelling Procedure

“…On average, the medians of AMS dx are 87% higher than the medians of AMS dm . The use of daily maximum data is inevitable in the presented study, although, in most previous flooding related studies [21,26,28], daily mean data were used to run the HT model. Secondly, the flood process typology was reviewed that also gives some indication of the length of flood events.…”

“…3. The event definition (i.e., selection of τ) may also depend on the intended application of the model, such as reconstruction of flood situations in the past lasting a certain number of days or application in a distinct (re)insurance context [21].…”

“…In a first step, the information about the intensity of these events was made available in real scale i.e., m 3 s −1 ) for each individual event and every station. To enable a further usage of the synthetic events, it was necessary to define the so-called level of extremeness [21] for all events and every single station on a probability scale. This probability scale corresponds to the concept of return period of hydraulic loads.…”